Discovery of action patterns in task-oriented learning processes

In this study, in order to support and facilitate the web-based learning, we concentrate on user learning behavior pattern discovery in a task-oriented learning process. Based on a hierarchical graph model which can describe relations among learning actions, learning activities, learning sub-tasks and learning tasks, we introduce the formal definitions for Learning Action Pattern and Goal-driven Learning Group to discover and represent users' learning behavior patterns within a learning task process. Two integrated algorithms are developed to calculate and generate the Learning Action Patterns for an individual user and the Goal-driven Learning Groups for a number of users, which can benefit sharing of learning activities and improve learning efficiency in e-learning environments. Finally, the design of a prototype system with experiment results is discussed.